A score that can accurately determine the risk of major bleeding during anticoagulant therapy may help to make decisions on anticoagulant use. RIETE is an ongoing registry of consecutive patients with acute venous thromboembolism (VTE). We composed a score to predict the risk for major bleeding within three months of anticoagulant therapy. Of 19,274 patients enrolled, 13,057 (67%) were randomly assigned to the derivation sample, 6,572 to the validation sample. In the derivation sample 314 (2.4%) patients bled (fatal bleeding, 105). On multivariate analysis, age >75 years, recent bleeding, cancer, creatinine levels >1.2 mg/dl, anemia, or pulmonary embolism at baseline were independently associated with an increased risk for major bleeding. A score was composed assigning 2 points to recent bleeding, 1.5 to abnormal creatinine levels or anemia, 1 point to the remaining variables. In the derivation sample 2,654 (20%) patients scored 0 points (low risk); 9,645 (74%) 1-4 points (intermediate); 758 (5.8%) >4 points (high risk). The incidences of major bleeding were: 0.3% (95% confidence interval [CI]: 0.1-0.6), 2.6% (95% CI: 2.3-2.9), and 7.3% (95% CI: 5.6-9.3), respectively. The likelihood ratio test was: 0.14 (95% CI: 0.07-0.27) for patients at low risk;2.96 (95% CI: 2.18-4.02) for those at high risk. In the validation sample the incidence of major bleeding was: 0.1%, 2.8%, and 6.2%, respectively. In conclusion, a risk score based on six variables documented at entry can identify VTE patients at low, intermediate, or high risk for major bleeding during the first three months of therapy.